Musk: OpenAI Betrayed the Promise of Safe AI

A. The Dangers of Artificial Intelligence

59. Over the course of the 20th century, the United States gradually shifted from a primarily labor-based economy to a knowledge-based one, with economic value increasingly generated primarily by human intelligence. As the century progressed, another paradigm shift was already underway: value creation through artificial intelligence (“AI”).

60. Starting in the late 2000s and early 2010s, an algorithm called “deep learning” was developed, the hallmark of which was that it no longer needed to be designed with significant knowledge of the task at hand because it could essentially “learn” from examples and program itself. As deep learning algorithms became increasingly sophisticated, some of the world’s leading AI researchers set their sights on Artificial General Intelligence (“AGI”). The basic concept of AGI is a general-purpose AI system—a machine having intelligence for a wide variety of tasks like a human.

61. Musk has long been concerned by the grave threat these advanced systems pose to humanity, which he has repeatedly warned is likely the greatest existential threat we face today. These dangers include, without limitation (or exaggeration), completely replacing the human workforce, supercharging the spread of disinformation, malicious human impersonation, and the manipulation of political and military systems, ultimately leading to the extinction of humanity. Musk’s concerns are shared by other leading figures before him like Stephen Hawking and Bill Joy who chillingly warned that with AGI, “the future doesn’t need us.”

62. Musk has publicly called for a variety of measures to address the dangers of AGI, from voluntary moratoria to regulation, but his calls largely fell on deaf ears.

63. Where some like Musk see AGI as an existential threat, others like Google—and as it would turn out, Defendants—see it as a source of even greater profit and power.

64. At the end of 2013, Musk learned that Google was planning to acquire DeepMind, which at the time was one of the most advanced AI companies in the industry. Musk, who is well-known for his opposition to closed technology—e.g., Musk’s rocket company SpaceX holds almost no patents and his electric vehicle company Tesla makes its patents open and available for public use—was deeply troubled by this development and believed that in the hands of a giant private company like Google, AGI would pose a particularly acute and noxious danger to humanity. To prevent this, Musk tried to stop the sale, but was ultimately unsuccessful.

65. In 2014, Google acquired DeepMind and with its team, was immediately catapulted to the front of the race for AGI.

66. Following Google’s acquisition, Musk began hosting a series of dinner discussions on ways to counter Google and promote AI safety. He even reached out to President Barack Obama in 2015 to discuss the issue, but regulation never came.

67. Musk continued to advocate for safe AI practices and in 2015, he thought he found someone who understood his concerns: Sam Altman.